Online personalization is the art of translating the human capacity of observation and “people watching” into the digital world. Instead of seeing with our eyes, we view human behavior through a stream of data, graphs and tables. Instead of offering a recommendation with our voice or our hands, we set personalization logic. The better we can codify the right logic to target the right visitor, the more successful we will be.
That’s why it’s very important to consider some of your assumptions about personalization. Is the data really telling you what you think it is? Can you completely explain behavior that you observe on a site? Are you really engaging your users, or are you just getting them to buy? In this post, I will talk about the top three pitfalls of personalization, and how to address them.
1. Assuming a Look Means a Desire
I once looked at a shirt printed with a goofy expression on a major clothing retailer’s site. I didn’t intend to purchase it, and in fact I was mostly interested in sharing it with some friends that would laugh at the message. Soon after, I noticed that every “recommended” area on the site was suddenly showing me products that were tangentially related to what I had viewed in almost every way. I even saw banner promotions for the manufacturer of the shirt, as well as the genre of the message I had viewed. All told, it seemed that about 30% of my viewable screen on this site was taken up with personalization around this one item.
It’s easy to fall into this trap. It makes sense, right? I look at pants, and you show me many pants. I look at blue pants, and you show me blue pants, and perhaps some blue shorts mixed in. The problem with this kind of logic is that it assumes that I want to actually purchase one of those products.
The key to solving this is to closely tie those that know your product best into the personalization program. In some cases, this might not be someone in the digital space, but instead a clothing designer or product manager that is more accustomed to blueprints than websites. Regardless of their background, they can offer very strong insights into the question of why someone is choosing a given product, and what that can mean for their next interest. To use the original example, that product manager may identify that novelty shirt as just that – a novelty, and recommend that I be shown some more comical options. While that may still not entice me to actually buy, it’s definitely more likely to get my attention.
2. Viewing a Single Page
A page is a step on a path. The incoming context and outgoing context sometimes matter far more than what is on a given page. However, all too often, retailers look at metrics and ask themselves why conversion is so low for a single page. Sometimes, this question can be the focus of a quarter or two of effort, and indeed sometimes that may be money well spent. But equally as often, the focus on a single page negates considering the makeup of traffic to and from it.
Moving through a site is a journey, not a series of moments. When a retailer views a report that says the traffic through the search page converts at a lower rate, they want to “improve conversion on the search page.” But if we parse that traffic down into source segments, we will find a great number of visitors that are buying at a high rate. The conversion rate is being dragged down by some traffic that doesn’t convert at all.
In our practice at Corra, we tend to look at groups of pages as a single target of opportunity. For example, we might test visitors that look at shirts, then perform a shirt-related keyword search, then hit a product page with a shirt, then add to cart. By looking at the whole path, and doing so in context of the key metric you want to improve, you can begin to account for various sources and destinations across your visitors.
3. Assuming Conversion Means Engagement
Let’s start with an example. Groupon likely has a very high conversion rate, and one could make the argument that they do reach a certain set of deal hunters in a profound way. But for most consumers, Groupon is a site that simply acts as a vehicle for getting what they want at a reasonable price – whether it’s clothing, meals, or experiences. Groupon doesn’t reach those customers because their true focus isn’t price, and yet that is almost Groupon’s entire model.
Likewise, we see many other sites go into a “promotion spiral” to drive up conversion. This is making a faulty assumption that a purchase equals a customer. In this case it doesn’t, because it only takes a good deal to get a single purchase. A subsequent full-price purchase would better put that customer on the path to being “engaged.”
So how do we engage without looking at conversion? The answer is to look for indicators that someone is engaged, and use that as a campaign metric. A very common metric is to look at subsequent return visits, and attribute that to a full engagement. Some retailers go further, and look at full-price purchases in conjunction with interacting with the site on more than one occasion.
A Final Tip
This is by no means a complete list, but it definitely captures the common pitfalls of personalization. I will close by sharing a tip that has served me well, though it tends to get odd looks at times. If you ever need a “gut check” on some personalization logic, remember that personalization is always aimed at targeting people. Try talking to the people around you and ask them what they think of seeing one product over another. Even if you aren’t talking to a digital marketing professional with 20 years of experience, you will get some great feedback that might even take you in some new directions.
Want to learn more about online personalization? Email email@example.com to get started!